Update MD034/no-bare-urls to better handle multiple similar issues on the same line (fixes https://github.com/igorshubovych/markdownlint-cli/issues/339).

demo/markdownlint-browser.js

@@ -47,7 +47,8 @@ module.exports.inlineCommentStartRe = inlineCommentStartRe;
 const htmlElementRe = /<(([A-Za-z][A-Za-z0-9-]*)(?:\s[^`>]*)?)\/?>/g;
 module.exports.htmlElementRe = htmlElementRe;
 // Regular expressions for range matching
-module.exports.bareUrlRe = /(?:http|ftp)s?:\/\/[^\s\]"']*(?:\/|[^\s\]"'\W])/ig;
+module.exports.bareUrlRe =
+    /(?:http|ftp)s?:\/\/[^\s\]<>"'`]*(?:\/|[^\s\]<>"'`\W])/ig;
 module.exports.listItemMarkerRe = /^([\s>]*)(?:[*+-]|\d+[.)])\s+/;
 module.exports.orderedListItemMarkerRe = /^[\s>]*0*(\d+)[.)]/;
 // Regular expression for all instances of emphasis markers
@@ -414,17 +415,22 @@ module.exports.flattenLists = function flattenLists(tokens) {
     }
     return flattenedLists;
 };
-// Calls the provided function for each specified inline child token
+/**
-module.exports.forEachInlineChild =
+ * Calls the provided function for each specified inline child token.
-    function forEachInlineChild(params, type, handler) {
+ *
-        filterTokens(params, "inline", function forToken(token) {
+ * @param {Object} params RuleParams instance.
-            for (const child of token.children) {
+ * @param {string} type Token type identifier.
-                if (child.type === type) {
+ * @param {Function} handler Callback function.
-                    handler(child, token);
+ * @returns {void}
-                }
+ */
-            }
+function forEachInlineChild(params, type, handler) {
-        });
+    filterTokens(params, "inline", (token) => {
-    };
+        for (const child of token.children.filter((c) => c.type === type)) {
+            handler(child, token);
+        }
+    });
+}
+module.exports.forEachInlineChild = forEachInlineChild;
 // Calls the provided function for each heading's content
 module.exports.forEachHeading = function forEachHeading(params, handler) {
     let heading = null;
@@ -585,6 +591,7 @@ module.exports.codeBlockAndSpanRanges = (params, lineMetadata) => {
  */
 module.exports.htmlElementRanges = (params, lineMetadata) => {
     const exclusions = [];
+    // Match with htmlElementRe
     forEachLine(lineMetadata, (line, lineIndex, inCode) => {
         let match = null;
         // eslint-disable-next-line no-unmodified-loop-condition
@@ -592,6 +599,32 @@ module.exports.htmlElementRanges = (params, lineMetadata) => {
             exclusions.push([lineIndex, match.index, match[0].length]);
         }
     });
+    // Match with html_inline
+    forEachInlineChild(params, "html_inline", (token, parent) => {
+        const parentContent = parent.content;
+        let tokenContent = token.content;
+        const parentIndex = parentContent.indexOf(tokenContent);
+        let deltaLines = 0;
+        let indent = 0;
+        for (let i = parentIndex - 1; i >= 0; i--) {
+            if (parentContent[i] === "\n") {
+                deltaLines++;
+            }
+            else if (deltaLines === 0) {
+                indent++;
+            }
+        }
+        let lineIndex = token.lineNumber - 1 + deltaLines;
+        do {
+            const index = tokenContent.indexOf("\n");
+            const length = (index === -1) ? tokenContent.length : index;
+            exclusions.push([lineIndex, indent, length]);
+            tokenContent = tokenContent.slice(length + 1);
+            lineIndex++;
+            indent = 0;
+        } while (tokenContent.length > 0);
+    });
+    // Return results
     return exclusions;
 };
 /**
@@ -3704,60 +3737,62 @@ module.exports = {
 "use strict";
 // @ts-check
 
-const { addErrorContext, bareUrlRe, filterTokens } = __webpack_require__(/*! ../helpers */ "../helpers/helpers.js");
+const { addErrorContext, bareUrlRe, withinAnyRange } = __webpack_require__(/*! ../helpers */ "../helpers/helpers.js");
-const htmlLinkOpenRe = /^<a[\s>]/i;
+const { codeBlockAndSpanRanges, htmlElementRanges, referenceLinkImageData } = __webpack_require__(/*! ./cache */ "../lib/cache.js");
-const htmlLinkCloseRe = /^<\/a[\s>]/i;
+const htmlLinkRe = /<a(?:|\s[^>]+)>[^<>]*<\/a\s*>/ig;
 module.exports = {
     "names": ["MD034", "no-bare-urls"],
     "description": "Bare URL used",
     "tags": ["links", "url"],
     "function": function MD034(params, onError) {
-        filterTokens(params, "inline", (token) => {
+        const { lines } = params;
-            let inLink = false;
+        const codeExclusions = [
-            let inInline = false;
+            ...codeBlockAndSpanRanges(),
-            for (const child of token.children) {
+            ...htmlElementRanges()
-                const { content, line, lineNumber, type } = child;
+        ];
+        const { definitionLineIndices } = referenceLinkImageData();
+        for (const [lineIndex, line] of lines.entries()) {
+            if (definitionLineIndices[0] === lineIndex) {
+                definitionLineIndices.shift();
+            }
+            else {
                 let match = null;
-                if (type === "link_open") {
+                const lineExclusions = [];
-                    inLink = true;
+                while ((match = htmlLinkRe.exec(line)) !== null) {
+                    lineExclusions.push([lineIndex, match.index, match[0].length]);
                 }
-                else if (type === "link_close") {
+                while ((match = bareUrlRe.exec(line)) !== null) {
-                    inLink = false;
+                    const [bareUrl] = match;
-                }
+                    const matchIndex = match.index;
-                else if ((type === "html_inline") && htmlLinkOpenRe.test(content)) {
+                    const bareUrlLength = bareUrl.length;
-                    inInline = true;
+                    const prefix = line.slice(0, matchIndex);
-                }
+                    const postfix = line.slice(matchIndex + bareUrlLength);
-                else if ((type === "html_inline") && htmlLinkCloseRe.test(content)) {
+                    if (
-                    inInline = false;
+                    // Allow ](... to avoid reporting Markdown-style links
-                }
+                    !(/\]\(\s*$/.test(prefix)) &&
-                else if ((type === "text") && !inLink && !inInline) {
+                        // Allow <...> to avoid reporting non-bare links
-                    while ((match = bareUrlRe.exec(content)) !== null) {
+                        !(prefix.endsWith("<") && /^[#)]?>/.test(postfix)) &&
-                        const [bareUrl] = match;
+                        // Allow [...] to avoid MD011/no-reversed-links and nested links
-                        const matchIndex = match.index;
+                        !(/\[[^\]]*$/.test(prefix) && /^[^[]*\]/.test(postfix)) &&
-                        const bareUrlLength = bareUrl.length;
+                        // Allow "..." and '...' for deliberately including a bare link
-                        // Allow "[LINK]" to avoid conflicts with MD011/no-reversed-links
+                        !(prefix.endsWith("\"") && postfix.startsWith("\"")) &&
-                        // Allow quoting as a way of deliberately including a bare URL
+                        !(prefix.endsWith("'") && postfix.startsWith("'")) &&
-                        const leftChar = content[matchIndex - 1];
+                        !withinAnyRange(lineExclusions, lineIndex, matchIndex, bareUrlLength) &&
-                        const rightChar = content[matchIndex + bareUrlLength];
+                        !withinAnyRange(codeExclusions, lineIndex, matchIndex, bareUrlLength)) {
-                        if (!((leftChar === "[") && (rightChar === "]")) &&
+                        const range = [
-                            !((leftChar === "\"") && (rightChar === "\"")) &&
+                            matchIndex + 1,
-                            !((leftChar === "'") && (rightChar === "'"))) {
+                            bareUrlLength
-                            const index = line.indexOf(content);
+                        ];
-                            const range = (index === -1) ? null : [
+                        const fixInfo = {
-                                index + matchIndex + 1,
+                            "editColumn": range[0],
-                                bareUrlLength
+                            "deleteCount": range[1],
-                            ];
+                            "insertText": `<${bareUrl}>`
-                            const fixInfo = range ? {
+                        };
-                                "editColumn": range[0],
+                        addErrorContext(onError, lineIndex + 1, bareUrl, null, null, range, fixInfo);
-                                "deleteCount": range[1],
-                                "insertText": `<${bareUrl}>`
-                            } : null;
-                            addErrorContext(onError, lineNumber, bareUrl, null, null, range, fixInfo);
-                        }
                     }
                 }
             }
-        });
+        }
     }
 };
 

helpers/helpers.js

@@ -23,7 +23,8 @@ const htmlElementRe = /<(([A-Za-z][A-Za-z0-9-]*)(?:\s[^`>]*)?)\/?>/g;
 module.exports.htmlElementRe = htmlElementRe;
 
 // Regular expressions for range matching
-module.exports.bareUrlRe = /(?:http|ftp)s?:\/\/[^\s\]"']*(?:\/|[^\s\]"'\W])/ig;
+module.exports.bareUrlRe =
+  /(?:http|ftp)s?:\/\/[^\s\]<>"'`]*(?:\/|[^\s\]<>"'`\W])/ig;
 module.exports.listItemMarkerRe = /^([\s>]*)(?:[*+-]|\d+[.)])\s+/;
 module.exports.orderedListItemMarkerRe = /^[\s>]*0*(\d+)[.)]/;
 
@@ -418,17 +419,22 @@ module.exports.flattenLists = function flattenLists(tokens) {
   return flattenedLists;
 };
 
-// Calls the provided function for each specified inline child token
+/**
-module.exports.forEachInlineChild =
+ * Calls the provided function for each specified inline child token.
+ *
+ * @param {Object} params RuleParams instance.
+ * @param {string} type Token type identifier.
+ * @param {Function} handler Callback function.
+ * @returns {void}
+ */
 function forEachInlineChild(params, type, handler) {
-  filterTokens(params, "inline", function forToken(token) {
+  filterTokens(params, "inline", (token) => {
-    for (const child of token.children) {
+    for (const child of token.children.filter((c) => c.type === type)) {
-      if (child.type === type) {
+      handler(child, token);
-        handler(child, token);
-      }
     }
   });
-};
+}
+module.exports.forEachInlineChild = forEachInlineChild;
 
 // Calls the provided function for each heading's content
 module.exports.forEachHeading = function forEachHeading(params, handler) {
@@ -608,6 +614,7 @@ module.exports.codeBlockAndSpanRanges = (params, lineMetadata) => {
  */
 module.exports.htmlElementRanges = (params, lineMetadata) => {
   const exclusions = [];
+  // Match with htmlElementRe
   forEachLine(lineMetadata, (line, lineIndex, inCode) => {
     let match = null;
     // eslint-disable-next-line no-unmodified-loop-condition
@@ -615,6 +622,31 @@ module.exports.htmlElementRanges = (params, lineMetadata) => {
       exclusions.push([ lineIndex, match.index, match[0].length ]);
     }
   });
+  // Match with html_inline
+  forEachInlineChild(params, "html_inline", (token, parent) => {
+    const parentContent = parent.content;
+    let tokenContent = token.content;
+    const parentIndex = parentContent.indexOf(tokenContent);
+    let deltaLines = 0;
+    let indent = 0;
+    for (let i = parentIndex - 1; i >= 0; i--) {
+      if (parentContent[i] === "\n") {
+        deltaLines++;
+      } else if (deltaLines === 0) {
+        indent++;
+      }
+    }
+    let lineIndex = token.lineNumber - 1 + deltaLines;
+    do {
+      const index = tokenContent.indexOf("\n");
+      const length = (index === -1) ? tokenContent.length : index;
+      exclusions.push([ lineIndex, indent, length ]);
+      tokenContent = tokenContent.slice(length + 1);
+      lineIndex++;
+      indent = 0;
+    } while (tokenContent.length > 0);
+  });
+  // Return results
   return exclusions;
 };
 

lib/md034.js

@@ -2,67 +2,76 @@
 
 "use strict";
 
-const { addErrorContext, bareUrlRe, filterTokens } = require("../helpers");
+const { addErrorContext, bareUrlRe, withinAnyRange } = require("../helpers");
+const { codeBlockAndSpanRanges, htmlElementRanges, referenceLinkImageData } =
+  require("./cache");
 
-const htmlLinkOpenRe = /^<a[\s>]/i;
+const htmlLinkRe = /<a(?:|\s[^>]+)>[^<>]*<\/a\s*>/ig;
-const htmlLinkCloseRe = /^<\/a[\s>]/i;
 
 module.exports = {
   "names": [ "MD034", "no-bare-urls" ],
   "description": "Bare URL used",
   "tags": [ "links", "url" ],
   "function": function MD034(params, onError) {
-    filterTokens(params, "inline", (token) => {
+    const { lines } = params;
-      let inLink = false;
+    const codeExclusions = [
-      let inInline = false;
+      ...codeBlockAndSpanRanges(),
-      for (const child of token.children) {
+      ...htmlElementRanges()
-        const { content, line, lineNumber, type } = child;
+    ];
+    const { definitionLineIndices } = referenceLinkImageData();
+    for (const [ lineIndex, line ] of lines.entries()) {
+      if (definitionLineIndices[0] === lineIndex) {
+        definitionLineIndices.shift();
+      } else {
         let match = null;
-        if (type === "link_open") {
+        const lineExclusions = [];
-          inLink = true;
+        while ((match = htmlLinkRe.exec(line)) !== null) {
-        } else if (type === "link_close") {
+          lineExclusions.push([ lineIndex, match.index, match[0].length ]);
-          inLink = false;
+        }
-        } else if ((type === "html_inline") && htmlLinkOpenRe.test(content)) {
+        while ((match = bareUrlRe.exec(line)) !== null) {
-          inInline = true;
+          const [ bareUrl ] = match;
-        } else if ((type === "html_inline") && htmlLinkCloseRe.test(content)) {
+          const matchIndex = match.index;
-          inInline = false;
+          const bareUrlLength = bareUrl.length;
-        } else if ((type === "text") && !inLink && !inInline) {
+          const prefix = line.slice(0, matchIndex);
-          while ((match = bareUrlRe.exec(content)) !== null) {
+          const postfix = line.slice(matchIndex + bareUrlLength);
-            const [ bareUrl ] = match;
+          if (
-            const matchIndex = match.index;
+            // Allow ](... to avoid reporting Markdown-style links
-            const bareUrlLength = bareUrl.length;
+            !(/\]\(\s*$/.test(prefix)) &&
-            // Allow "[LINK]" to avoid conflicts with MD011/no-reversed-links
+            // Allow <...> to avoid reporting non-bare links
-            // Allow quoting as a way of deliberately including a bare URL
+            !(prefix.endsWith("<") && /^[#)]?>/.test(postfix)) &&
-            const leftChar = content[matchIndex - 1];
+            // Allow [...] to avoid MD011/no-reversed-links and nested links
-            const rightChar = content[matchIndex + bareUrlLength];
+            !(/\[[^\]]*$/.test(prefix) && /^[^[]*\]/.test(postfix)) &&
-            if (
+            // Allow "..." and '...' for deliberately including a bare link
-              !((leftChar === "[") && (rightChar === "]")) &&
+            !(prefix.endsWith("\"") && postfix.startsWith("\"")) &&
-              !((leftChar === "\"") && (rightChar === "\"")) &&
+            !(prefix.endsWith("'") && postfix.startsWith("'")) &&
-              !((leftChar === "'") && (rightChar === "'"))
+            !withinAnyRange(
-            ) {
+              lineExclusions, lineIndex, matchIndex, bareUrlLength
-              const index = line.indexOf(content);
+            ) &&
-              const range = (index === -1) ? null : [
+            !withinAnyRange(
-                index + matchIndex + 1,
+              codeExclusions, lineIndex, matchIndex, bareUrlLength
-                bareUrlLength
+            )
-              ];
+          ) {
-              const fixInfo = range ? {
+            const range = [
-                "editColumn": range[0],
+              matchIndex + 1,
-                "deleteCount": range[1],
+              bareUrlLength
-                "insertText": `<${bareUrl}>`
+            ];
-              } : null;
+            const fixInfo = {
-              addErrorContext(
+              "editColumn": range[0],
-                onError,
+              "deleteCount": range[1],
-                lineNumber,
+              "insertText": `<${bareUrl}>`
-                bareUrl,
+            };
-                null,
+            addErrorContext(
-                null,
+              onError,
-                range,
+              lineIndex + 1,
-                fixInfo
+              bareUrl,
-              );
+              null,
-            }
+              null,
+              range,
+              fixInfo
+            );
           }
         }
       }
-    });
+    }
   }
 };

test/bare-urls.md

@@ -29,3 +29,21 @@ As is <a href="https://example.com/info.htm">https://example.com/info.htm text</
 <br> Another violation: https://example.com. {MD034} <br>
 
 <br/> Another violation: https://example.com. {MD034} <br/>
+
+This is not a bare [link]( https://example.com ).
+
+URLs in HTML are not bare:
+
+<element-name first-attribute=" https://example.com/first " second-attribute=" https://example.com/second ">
+  Text
+</element-name>
+
+<element-name
+  first-attribute=" https://example.com/first "
+  second-attribute=" https://example.com/second "></element-name>
+
+URLs in link and image text are not bare:
+
+Text [link to https://example.com site](https://example.com) text.
+
+Image ![for https://example.com site](https://example.com) text.

test/links.md

@@ -28,3 +28,9 @@ Other enclosures are not allowed:
 
 (https://example.com) {MD034}
 {https://example.com} {MD034}
+
+Duplicate links in tables should be handled:
+
+| Link                 | Same Link            | Violation |
+|----------------------|----------------------|-----------|
+| https://example.com/ | https://example.com/ | {MD034}   |

test/snapshots/markdownlint-test-scenarios.js.md

@@ -2998,6 +2998,24 @@ Generated by [AVA](https://avajs.dev).
       <br> Another violation: <https://example.com>. {MD034} <br>␊
       ␊
       <br/> Another violation: <https://example.com>. {MD034} <br/>␊
+      ␊
+      This is not a bare [link]( https://example.com ).␊
+      ␊
+      URLs in HTML are not bare:␊
+      ␊
+      <element-name first-attribute=" https://example.com/first " second-attribute=" https://example.com/second ">␊
+        Text␊
+      </element-name>␊
+      ␊
+      <element-name␊
+        first-attribute=" https://example.com/first "␊
+        second-attribute=" https://example.com/second "></element-name>␊
+      ␊
+      URLs in link and image text are not bare:␊
+      ␊
+      Text [link to https://example.com site](https://example.com) text.␊
+      ␊
+      Image ![for https://example.com site](https://example.com) text.␊
       `,
     }
 
@@ -21928,8 +21946,15 @@ Generated by [AVA](https://avajs.dev).
         {
           errorContext: 'https://example.com/same',
           errorDetail: null,
-          errorRange: null,
+          errorRange: [
-          fixInfo: null,
+            46,
+            24,
+          ],
+          fixInfo: {
+            deleteCount: 24,
+            editColumn: 46,
+            insertText: '<https://example.com/same>',
+          },
           lineNumber: 26,
           ruleDescription: 'Bare URL used',
           ruleInformation: 'https://github.com/DavidAnson/markdownlint/blob/v0.0.0/doc/md034.md',
@@ -22144,7 +22169,7 @@ Generated by [AVA](https://avajs.dev).
       ␊
       Text <https://example.com/same> more text <https://example.com/same> still more text <https://example.com/same> done {MD034}␊
       ␊
-      Text <https://example.com/same> more \\* text https://example.com/same more \\[ text <https://example.com/same> done {MD034}␊
+      Text <https://example.com/same> more \\* text <https://example.com/same> more \\[ text <https://example.com/same> done {MD034}␊
       ␊
       Text <https://example.com/first> more text <https://example.com/second> still more text <https://example.com/third> done {MD034}␊
       ␊
@@ -22379,6 +22404,46 @@ Generated by [AVA](https://avajs.dev).
             'no-bare-urls',
           ],
         },
+        {
+          errorContext: 'https://example.com/',
+          errorDetail: null,
+          errorRange: [
+            3,
+            20,
+          ],
+          fixInfo: {
+            deleteCount: 20,
+            editColumn: 3,
+            insertText: '<https://example.com/>',
+          },
+          lineNumber: 36,
+          ruleDescription: 'Bare URL used',
+          ruleInformation: 'https://github.com/DavidAnson/markdownlint/blob/v0.0.0/doc/md034.md',
+          ruleNames: [
+            'MD034',
+            'no-bare-urls',
+          ],
+        },
+        {
+          errorContext: 'https://example.com/',
+          errorDetail: null,
+          errorRange: [
+            26,
+            20,
+          ],
+          fixInfo: {
+            deleteCount: 20,
+            editColumn: 26,
+            insertText: '<https://example.com/>',
+          },
+          lineNumber: 36,
+          ruleDescription: 'Bare URL used',
+          ruleInformation: 'https://github.com/DavidAnson/markdownlint/blob/v0.0.0/doc/md034.md',
+          ruleNames: [
+            'MD034',
+            'no-bare-urls',
+          ],
+        },
       ],
       fixed: `# Link test␊
       ␊
@@ -22410,6 +22475,12 @@ Generated by [AVA](https://avajs.dev).
       ␊
       (<https://example.com>) {MD034}␊
       {<https://example.com>} {MD034}␊
+      ␊
+      Duplicate links in tables should be handled:␊
+      ␊
+      | Link                 | Same Link            | Violation |␊
+      |----------------------|----------------------|-----------|␊
+      | <https://example.com/> | <https://example.com/> | {MD034}   |␊
       `,
     }